The Agent Skills Directory

[SAFE]: All scripts are implemented using only the Python standard library, with no requirements for third-party packages.
[EXTERNAL_DOWNLOADS]: The documentation provides instructions for manual installation via git clone from a GitHub repository.
[COMMAND_EXECUTION]: The skill includes scripts for capturing and recalling memories that perform local file I/O within the workspace. These scripts do not execute shell commands or process user input as executable code.
[PROMPT_INJECTION]: The memory protocol instructions in SKILL.md guide the agent's use of the memory tools without attempting to bypass safety filters or override core instructions.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection through the retrieval of conversation context.
Ingestion points: recall.py reads content from MEMORY.md and daily log files in the memory/ directory.
Boundary markers: Absent; memory snippets are provided to the agent without specific delimiters or instructions to ignore embedded commands.
Capability inventory: Local file-write access via capture.py and consolidate.py. No network or subprocess execution capabilities exist.
Sanitization: Content retrieved from memory is not sanitized or escaped before being presented to the agent.

memory